Model-based atrial fibrillation detection

Abstract

Atrial fibrillation (AF) is the most common cardiac arrhythmia, leading to several patient risks. This kind of arrhythmia affects mostly elderly people, in particular those who suffer from heart failure disease, one of the main causes of hospitalization. Thus, detection of AF becomes decisive in the diagnosis and prevention of cardiac threats, with particular interest in the context of pHealth solutions.This chapter presents a real-time AF detection scheme, developed within the MyHeart project, a pHealth project financed by the European Commission. The proposed strategy is based on a computational intelligence approach, combining expert knowledge and neural networks. In particular, it makes use of the three principal physiological characteristics of AF, applied by cardiologists in their daily reasoning: P wave absence/presence, heart rate irregularity, and atrial activity analysis. This knowledge-based approach has the advantage of increasing interpretability of the results to the medical community, while improving detection robustness.The clinical validation of the strategy is performed using public databases (MIT-BIH Arrhythmia and QT databases from Physionet) as well as the ECG data acquired with the MyHeart vest during the MyHeart trial.